4 ms·

Composites can be tested. Not as easily as metals can be, and it involves a lot more software analysis and even AI, but it's still possible. I work on flaw dete

Composites can be tested. Not as easily as metals can be, and it involves a lot more software analysis and even AI, but it's still possible. I work on flaw detection/segmentation with ultrasounds -and I have not directly worked with metal scans so I can't exactly compare- but we can already get some pretty good results with the right probes and scanners. The other major issue is that NDT on composites needs more care and time from the inspectors on the ground, and might need a lot more organizational resources to track and make sense of the data. But now that i think about it, I have no clue about performance or even feasibility of testing carbon composites. I'll have to ask around, but I'm pretty sure we have clients that use our hardware for Carbon! :)

jacquesm 3y ago
How does that work? Shouldn't the device be certified for such an application?
- mardifoufs 3y ago
  Yes! Sorry if I wasn't clear enough. I meant that they use the devices because they are compliant and certified for carbon composites. I just wasn't sure since the devices are usually not for a single purpose so it's hard to keep track of all certified use cases from my PoV. But sometimes they are! We have a few scanners specifically for pipelines and even robots for wind blades, yet even then the probe usually uses the same detection technology (ultrasounds, eddy currents, etc). The magic happens on the signal processing side when it comes to the difference between the scanners, and the trend is towards multipurpose all in ones. Our most recent industrial scanner has an FPGA that is explicitly there to allow for extensibility so that it fits most applications. FPGAs have been used quite often in the field but usually merely to implement some specific signal processing, not to be extensible! I'm rambling here but my point is that some devices can be used for pretty much any application (the Omniscan for example), process specific devices are more for large scale industrial workflows! The probes can be the only thing you need to change to use the same scanner to do stuff as varied as hydrogen poisoning analysis on steel to composite flaw detection on windblades! Here is an example, again with the Omniscan : https://www.olympus-ims.com/en/scanners/aerospace-wind-blade-inspection-scanners/ https://www.olympus-ims.com/en/scanners/aerospace-wind-blade...
  - jacquesm 3y ago
    Super interesting work, thank you for the follow up. Some of my customers use stuff like this so it's nice to see someone from the supply side. It's interesting to think about the different kinds of defects. "Not like the rest" is a fairly easy kind of defect to detect, "Structurally different from the baseline" is a bit harder and a whole batch that is different in some important aspect is quite hard. FPGAs are super powerful for applications like these, I see them used in plenty of sensors where flexibility and being future proof is a must so that you don't need to redesign the entire sensor for every variation on the theme.

jacquesm 3y ago

How does that work? Shouldn't the device be certified for such an application?

mardifoufs 3y ago
Yes! Sorry if I wasn't clear enough. I meant that they use the devices because they are compliant and certified for carbon composites. I just wasn't sure since the devices are usually not for a single purpose so it's hard to keep track of all certified use cases from my PoV. But sometimes they are! We have a few scanners specifically for pipelines and even robots for wind blades, yet even then the probe usually uses the same detection technology (ultrasounds, eddy currents, etc). The magic happens on the signal processing side when it comes to the difference between the scanners, and the trend is towards multipurpose all in ones. Our most recent industrial scanner has an FPGA that is explicitly there to allow for extensibility so that it fits most applications. FPGAs have been used quite often in the field but usually merely to implement some specific signal processing, not to be extensible! I'm rambling here but my point is that some devices can be used for pretty much any application (the Omniscan for example), process specific devices are more for large scale industrial workflows! The probes can be the only thing you need to change to use the same scanner to do stuff as varied as hydrogen poisoning analysis on steel to composite flaw detection on windblades! Here is an example, again with the Omniscan : https://www.olympus-ims.com/en/scanners/aerospace-wind-blade-inspection-scanners/ https://www.olympus-ims.com/en/scanners/aerospace-wind-blade...
- jacquesm 3y ago
  Super interesting work, thank you for the follow up. Some of my customers use stuff like this so it's nice to see someone from the supply side. It's interesting to think about the different kinds of defects. "Not like the rest" is a fairly easy kind of defect to detect, "Structurally different from the baseline" is a bit harder and a whole batch that is different in some important aspect is quite hard. FPGAs are super powerful for applications like these, I see them used in plenty of sensors where flexibility and being future proof is a must so that you don't need to redesign the entire sensor for every variation on the theme.